JOURNAL OF MASS SPECTROMETRY, VOL. 31, 247-254 (1996)

Elucidation of Isomeric Structures for Ubiquitin [ M + 12H] l2 + Ions Produced by Electrospray Ionization Mass Spectrometry

Carolyn J. Cassady* and Scott R. Carr Department of Chemistry, Miami University, Oxford, OH 45056

Gas-phase deprotonation reactions, hydrogeAeuterium exchange reactions and collision-induced dissociation (CID) were used to distinguish between two isomeric forms of [ M + 12H] IZ+ produced from the protein ubiquitin. Ions were generated by electrospray ionization and studied in a Fourier transform ion cyclotron resonance mass spectrometer. For [ M + 12Hj IZ+ formed directly from the electrospray process, deprotonation reactions with ammonia and 2-fluoropyridine yield non-linear pseudo-first-order kinetic behavior that indicates the presence of two ion structures. The fraction of ions that undergo the fastest deprotonation reactions, and is presumably the least energetically stable isomer, accounts for - 60% of the [ M + 12H1 12+ produced by electrospray. In reactions with D,O and CD,OD, the [M + 12H)12+ which are deprotonated faster exchange the first 11 f 1 hydrogens more readily that the remaining [M + 12Hj"+ population. Results from CID experiments, obtained as a function of reaction time with the amines, also indicate the existence of more than one [ M + 12HJi2+ structure. The CID fragmentation patterns provide information about the general locations of the charge sites. Surprisingly, evidence for only one structure (the slow-reacting, more stable species) is found for [M + 12H]"+ that is produc- ed by gas-phase deprotonation of [ M + 13H] 13+, which is the 'fully protonated' form of ubiquitin. These results are discussed in terms of ubiquitin isomers related to protonation site and three-dimensional conformation.

KEYWORDS : electrospray ionization; Fourier transform ion cyclotron resonance mass spectrometry; ubiqui- tin ; collision-induced dissociation; deprotonation reactions

INTRODUCTION

The determination of gas-phase ion structures is an active area of research which increases our understand- ing of the processes that occur inside a mass spectro- meter and provides analytical information about molecular structures. Historically, this work has been limited to singly charged ions because these species are generated by common techniques such as electron impact (EI) ionization, chemical ionization (CI) and fast atom bombardment. The recent development of electro- spray ionization mass spectrometry (EST-MS),'.' however, has made possible the formation of large multiply charged ions from samples such as protein^,^ carbohydrate^.^ nucleotides5 and inorganic complexes.6

The large multiply charged ions produced by ESI have the potential for yielding many challenging ion structure problems. Several studies have shown that solution-phase conformation, as controlled by varying solution conditions, can affect the charge-state distribu- tions of ions generated by ESJ.'-'* Direct evidence for gas-phase conformational isomers of the protein cyto- chrome c has been obtained by McLafferty and co- workers using hydrogen-deuterium (H-D) exchange reactions performed in a Fourier transform ion cyclo- tron resonance (FT-ICR) mass spectrometer.' 'J Gas- phase conformers of protein ions have also been

Author to whom correspondence should be addressed.

distinguished using their collision cross-sections.' '-" In addition, gas-phase deprotonation reaction studies in our laboratory have revealed non-linear pseudo-first- order kinetic behavior that suggests the presence of multiple structural isomers for ubiquitin [M + nH]"+, n = 4-6.16

The current investigation demonstrates that the same tandem mass spectrometric (MS/MS) techniques as uti- lized to elucidate isomeric structures of singly charged ions can be applied to multiply protonated ions, [M + nH]"', generated by ESI. Conclusive evidence for at least two structures of [M + 12H]'2f from the protein ubiquitin has been obtained by ion-molecule reactions and collision-induced dissociation (CID). In addition, it is shown that the [M + 12H]"+ popu- lation generated directly by ESI is not identical to the [M + 12HI1'+ population formed by gas-phase depro- tonation of [M + 13H]I3+.

EXPERIMENTAL

All experiments were performed using a Bruker (Billerica, MA, USA) CMS 47-X FT-ICR mass spectro- meter equipped with a 4.7 T superconducting magnet and an external ion source.17 Ions were produced using an Analytica of Branford (Branford, CT, USA) ESI source mounted in place of the normal EI/CI source of the FT-ICR. Experimental details regarding ion pro- duction by EST and ion transfer into the FT-ICR cell

CCC 1076-5 174/96/030247-08 0 1996 by John Wiley & Sons, Ltd.

Received 22 July 1995 Accepted 6 November 1995

248 C. J. CASSADY A N D S. R. CARR

have been discussed previously.16 To minimize ion internal energy, the nozzle-skimmer bias of the ESI source was < 80 V, and the FT-ICR cell potentials were held constant at + 1.0 V.

Bovine ubiquitin was purchased from Fluka Chemi- cal (Ronkonkoma, NY, USA) and used without addi- tional purification. Ions were formed from a solution that contained ubiquitin at 1.2 x M (0.1 mg ml-') in methanol-water-acetic acid (49 : 49 : 2.0, v/v/v). In deprotonation reaction experiments, [M + nH]"+ were m/z-selected by resonant frequency ejection techniques" and allowed to react with static pressures of ammonia or 2-fluoropyridine in the range (1- 80) x lo-' Torr (1 Torr = 133.3 Pa); for example,

[M + 12H]I2+ + B + [M + l lH]ll+ + BH+

Bimolecular rate constants (kexp) were determined by observing the pseudo-first order change in reactant ion abundance as a function of time at a constant pressure. Since the image current produced by an ion during detection in FT-ICR is directly proportional to the ion's charge, all measured ion abundances were corrected for charge state prior to calculation of experimental rate constants. Pressures were measured with a calibrated ionization gaugeIg and corrected for the reactant gas ionization eficiency.20

Hydrogen-deuterium exchange reactions employed m/z-selected [M + nH]"+, n = 11-13, reacting with D 2 0 and with CD,OD. The pressures of the deuterated reagents were in the range (4-50) x Torr. Reaction times were varied from 0 to 120 s.

CID experiments utilized sustained off-resonance irradiation (SORI)" to induce low-energy fragmenta- tion. Ions were activated 500-1000 Hz off-resonance (either higher or lower frequency) with a 5-10 V,, pulse for 50-150 ms in the presence of a pulsed pressure of xenon that reached a maximum of Torr. The pulse sequence for CID experiments included : isolation of [M + 12H]I2+, a variable reaction delay, re-isolation of [M + 12H]12+ to remove unwanted reaction products, SORI ion activation, a delay to allow collisions and swept frequency excitation with image current detec- tion.

RESULTS

Ubiquitin (M, = 8564.8) is a monomeric peptide con- taining 76 amino acid residues.22 Twelve of the residues are basic in solution: four arginines, seven lysines and one histidine. For peptides, ESI preferentially proto- nates these sites and the N-terminal amino group.23 This corresponds to 13 basic sites, which is consistent with [M + 13H]I3+ being the most highly charged ubiquitin ion observed in our laboratory. The focus of the present study, [M + 12H]12+, is lacking one proton relative to 'fully' protonated ubiquitin.

Deprotonation reactions

We recently reported that [M + 12H]I2+ of ubiquitin were rapidly deprotonated by compounds having gas-

phase basicities (GB) greater than or equal to that of propylamine (GB = 210.1 kcal mol-' (1 kcal = 4.184 kJ)24).'6 These reactions exhibit linear pseudo-first- order kinetic behavior and provide no evidence for more than one [M + 12H]12' structure. In contrast, deprotonation reactions of ESI-generated [M + 12H]12+ with the less basic ammonia (GB = 195.6 kcal mol-') (Ref. 24) and 2-fluoropyridine (GB = 202.8 kcal mol-') (Ref. 24) reveal non-linear pseudo-first-order kinetic behavior that suggests the presence of at least two ion structures. For example, Fig. 1 gives semi-logarithmic plots of the relative inten- sity of [M + 12H]12+ us. reaction time with ammonia and 2-fluoropyridine; the data are markedly non-linear. These experiments were repeated several times over a period of 3 months; in all cases the non-linear kinetic behavior was observed. Voltage modifications at the ESI source and the ion extractions optics had no noticeable impact on the results. Interestingly, this behavior appears to be unique to n = 12, with n = 11 and 13 exhibiting only linear pseudo-first-order kinetic behavior.

By fitting the deprotonation reaction data to the sum of two exponential functions, reaction rate constants for two [M + 12H]"+ populations were obtained. With ammonia, 60(_+10)% of [M + 12H]"+ react with a rate constant (kexp) of 8.4(_+4.4) x lo-" cm3 molecule- ' s - ', while the remaining 40( & lo)% of the ions are slower to react with k,,, = 5.4(f2.0) x

I I I I , ,

0 10 20 30 40 50 60 E

Reaction Time, seconds

0

0 0

g 1.0

c 0.6 a n 4

c: 0.3 .- - I Pl

+ - f, 0.1 - m E 0 10 20 30 40 50 60

Reaction Time, seconds

Figure 1. Reactant loss curves for the reactions of ubiquitin [M + 12H]'2+ generated directly from ESI with (a) ammonia at 2.0 x lo-' Torr and (b) 2-fluoropyridine at 3.9 x lo-' Torr. The logarithm of the abundance of [M + 12H]"+ (relative to the total abundance for [M +nH]"+ product and parent ions) is plotted as a function of reaction time. The fitted curve represents a non-linear regression of the data to the sum of two exponentials.

ISOMERIC STRUCTURES OF UBIQUITIN IONS PRODUCED BY ESI 249

cm3 molecule-' s- '. For 2-fluoropyridine the difference is more pronounced : fast-reacting ions account for 46(f5)% of the [M + 12H]12+ and have kexp = 3.8 (k1.5) x lo-'' cm3 molecule-'^-^, while 54(_+5)% of the ions react with kexp = 2.8( f 0.2) x lo-'' cm3 molecule-' s-'. (To distinguish between the two [M + 12H]12+ isomers in this paper, the ions will be referred to as 'fast-reacting' and 'slow-reacting' based on their relative rates of deprotonation.)

Average dipole orientation theoryz5 yields a collision rate constant of (1-2) x lo-' cm3 molecule-' s- ' for these reactions. Given that the experimental rate con- stants are 10z-105 times lower than the collision rate constants, these processes are very inefficient and pre- sumably endoergic. The fast-reacting ions are deproton- ated by 2-fluoropyridine at a rate that is 10' times greater than the other rates discussed above. Therefore, the lower percentage of fast-reacting ions found with 2- fluoropyridine (46% us. 60% from ammonia reactions) may result from selective depletion of these ions by de- protonation during the 100-200 ms period of ion accu- mulation in the FT-ICR cell. In other words, -60% is probably the more accurate percentage of fast-reacting [M + 12H]lZf generated by ESI.

In addition to production directly by ESI, [M + 12H]12+ were formed by removal of a proton from 'fully protonated [M + 13H]13+ with 2- fluoropyridine. Surprisingly, these ions show kinetic evi- dence for only one [M + 121'" population and react with a rate constant of 5.l(f0.8) x cm3 molecule- s - '. Therefore, only slow-reacting ions are produced in abundance by deprotonation of [M + 13H]13+. Since there was a possibility that the fast-reacting [M + 12H]"+ produced from [M + 13H]13+ reacted away prior to the isolation of [M + 12H]12+ for reactivity studies, a variety of pres- sures and time-scales were used in these experiments.

However, no evidence for production of the fast- reacting n = 12 population from n = 13 was found. Assuming that the deprotonation reactions are more energetically favorable for fast-reacting [M + 12H] l2 +, then these ions are the least stable; in turn, the slow- reacting ions are more stable. This suggests that depro- tonation of [M + 13H]I3+, which yields exclusively slow-reacting [M + 12H]"+, produces a more homo- geneous and energetically stable ion population than the ESI process itself. Thus, ESI is less selective than ion/molecule reactions at protonating (or deprotonating) basic sites on proteins.

Hydrogen-deuterium exchange reactions

Ubiquitin [M + nH]"+, n = 11-13, react with CD30D and D,O by exclusively H-D exchange. Around 40-50% of the [M + 12H]"+ ions exchange the first 11 ? 1 hydrogens more rapidly than the remaining [M + 12HI1'+. The result is a bimodal distribution such as that shown in Fig. 2(b) for [M + 12H]12+ reacting with 4 x Torr of CD30D for 10.0 s. Under these conditions, the ions split into two envelopes of peaks within a few seconds; the two envelopes then continue to exchange at fairly similar rates. At the longest time monitored, 120.0 s, approx- imately 30 exchanges have occurred and the spacing between the two envelopes is still about the same as that observed at 10.0 s. Owing to the slowness of the reactions, the maximum number of exchanges was not monitored.

In H-D exchange reactions, [M + 12H]"+ again exhibits reactive behavior that is not observed for its neighboring charge states. For example, Fig. 2(a) shows the reaction of the [M + llH]"+ distribution under

,I I [M + 12H112+

715.0 716.0 717.0 718.0 719.0 m/z Y l ~ j ~ l ~ ~ ~ , ~ ~ r , ~ l , l ,

710 730 750 770 m / Z

Figure 2. Mass spectra obtained when ESI-generated ubiquitin (a) [M + 11 H]ll + and (b) [M + 12H]''+ react by H-D exchange with 4 x lo-' Torr of CD,OD for 10.0 s. The insets are expansions of the regions around the parent ions. Before the exchange reactions, the parent ion distributions of were about 1 m/z wide owing to the ''C/''C isotopic peaks.

250 C. J. CASSADY AND S. R. CARR

80 - the same conditions of those in Fig. 2(b) for [M + 12H]12+. As H-D exchanges occur, the n = 11 distribution of isotopic peaks does not split into two envelopes; instead it broadens as the number of deute- riums incorporated into the ions increases. Even after 120.0 s, no splitting is observed. The [M + 13H]13+ system undergoes H-D exchanges in a manner similar to that of [M + 11H]"+. In comparing the number of exchanges at set times between the various parent ions, the fast-exchanging n = 12 ions react at roughly the same rate as the n = 11 and 13 populations. Thus, it is the slow-exchanging [M + 12HIi2+ that are unique.

In order to correlate the two [M + 12H]12 'distributions observed by deprotonation to the two populations found by H-D exchange, experi- ments were performed involving two consecutive reac- tions. During the first reaction period, the fast-deprotonating [M + 12H]12+ were selectively removed by reaction with ammonia. When the remain- ing [M + 12H]"+ were allowed to undergo H-D exchange with CD,OD, the distribution did not split into two envelopes of peaks. Instead, the remaining ions react at roughly the same rate as the population of [M + 12H]12+ that is slow to exchange. This suggests that the [M + 12H]12+ ions that are more slowly dep- rotonated are also the ions that are slow to H-D exchange.

100 - 80 -

60 -

40:

20 -

Collision-induced dissociation

Y

The presence of at least two different [M + 12H]"+ structures was confirmed by SORI CID experiments performed as a function of reaction time with ammonia and with 2-fluoropyridine. The reaction time, during which ubiquitin's fast-reacting [M + 12H]12+ is selec- tively depleted, occurs before SORI activation. Typical SORI CID spectra are given in Fig. 3 for ESI-generated [M + 12H]12+. Figure 3(a) involves no reaction time with 2-fluoropyridine and therefore depicts CID of all [M + 12H]"+. The spectrum in Fig. 3(b) involves dis- sociation of the slow-reacting [M + 12H]"+; under the experimental conditions employed, less than 0.02% of the fast-reacting isomer should remain. (This value was calculated using the reaction time, the 2- fluoropyridine pressure and kexp for the reaction of the fast-reacting isomer with 2-fluoropyridine.)

As the SORI CID spectra in Fig. 3 illustrate, a gradual shift in the relative abundances of several frag- ment ions occurs as the deprotonation reaction prog- resses. After the reaction has removed the fast-reacting [M + 12H]12+ population, this shift ceases and the CID spectra remain virtually unchanged as the reaction delay is increased. This is illustrated for the y5* product ions in Fig. 4. (The nomenclature used in Figs 3 and 4 and in the text to describe fragmentation is based on conventional peptide dissociation notation with a superscript added to show the charge of the fragment.26 The y ions are formed from the C-terminus end of the molecule and involve cleavage of an (O=C)N bond. Subscripts on y denote residue position for this cleavage as counted from the C-terminus.) As Fig. 4 shows, the first 1.0 s of the reaction time yields marked changes in

20

0

4+ Y 24

a+ Y5a

I

600 700 BOO 900 1000 1100 mlz

0 - L

5 + 0

10-

4+ y24

9 C Y 58

I- , . ' . ' ' I 1 , , , 1 , , , ! , , ' 1 . ' . - I ' '-7 500 600 700 800 900 I000 1100

mlz Figure 3. SORI CID mass spectra obtained from ESI-generated ubiquitin [M + 1 2H]12+ following reaction with 2-fluoropyridine at a static pressure of 2.3 x lo-' Torr for (a) 0.0 and (b) 3.0 s. Spectrum (a) involves both [M + 1 2HIi2+ populations. Spectrum (b) involves the slow-reacting isomer; under these experimental conditions, the fast-reacting isomer should account for less than 1%ofthe[M+12H]'2+.

the relative abundances of y!8+, y z l and y t i+ ; however, beyond 1.0 s the ratios remain relative stable. Under the conditions of Fig. 4, 94% of the fast-reacting [M + 12H]"+ are removed within the first 1.0 s.

I 1 I I 1 I

0 1 2 3 4 5 6

Reaction Time, seconds

Figure 4. Relative abundances of SORI CID y6* product ions as a function of reaction time. The parent ions, ESI-generated ubiquitin [M + 1 2H]i2+, are allowed to react with 2-fluoropryidine (2.3 x lo-' Torr) before collisional activation.

ISOMERIC STRUCTURES OF UBIQUITIN IONS PRODUCED BY ESI 251

Therefore, after longer reaction times only the slow- reacting [M + 12H]"+ remain for study by SORI CID. In addition, the CID spectra do not change as a function of reaction delay when the amine is replaced with argon. This is important because it indicates that collisional thermalization is not responsible for the phe- nomenon of changing product abundances as a function of reaction time. Also, the spectra in Fig. 3 were acquired within seconds of one another, with no experi- mental conditions being changed except for the length of this pre-activation delay.

In terms of SORI CID product ion abundances, the spectra for [M + 12H]"+ formed by deprotonation of [M + 13H]l3+ most closely resemble the spectra of the slow-reacting ions and are independent of reaction delay with 2-fluoropyridine or ammonia. This agrees with the deprotonation reaction results.

Sustained off-resonance irradiation (SORI) during CID in an FT-ICR was first employed on singly charged ions by Jacobson and c o - ~ o r k e r s . ~ ~ ~ ~ ' McLaf- ferty and co-workers28 have recently shown that SORI is more efficient at producing fragmentation of large multiply charged ions than the on-resonance irradiation that is commonly employed in FT-ICR CID experi- ments. Given that the application of SORI CID on multiply charged ions generated by ESI is relatively new, it is worthwhile to discuss our experimental condi- tions. As an example, for the spectra shown in Fig. 3, SORI activation involved irradiation of the trapped parent ions with a 5 V,, pulse for 120 ms at a frequency that was 500 Hz lower than the cyclotron frequency of [M + 12H]"+. Immediately before this SORI pulse, the collision gas xenon was admitted into the FT-ICR cell vacuum chamber to a peak pressure of lo-' Torr; the xenon was pumped away within 1000 ms. The SORI pulse was followed by a 1000 ms delay prior to ion detection. Parent ion dissociation occurred during both the 120 ms SORI pulse and the 1000 ms delay. Our experiments differ from other SORI ESI experiments2s929 in that the SORI pulse is closer in fre- quency to the parent ion and occurs for a shorter dura- tion. For example, in past studies SORI CID has been performed on ubiquitin [M + nH]"+ with an excitation pulse 2 kHz off-resonance administered for 1 or 2 s .~' Our shorter, 'harder' SORI conditions minimize the abundance of uninformative, low-energy water loss pro- cesses in the spectra. They also minimize the 'blind

which is a portion of the SORI CID spec- trum that contains no ions and results from ejection of product ions that are in close proximity to the irradia- tion frequency. A blind spot 40 m/z wide at m/z 1000 has been reported with a 2 kHz off-resonance pulse in an FT-ICR with a 6 T magnetic field." Since a decrease in magnetic field strength should increase the size of the blind spot, experiments in our 4.7 T FT-ICR might have an even larger blind spot under such conditions. However, the spectra shown in Fig. 3 have a blind spot of less than 5 m/z. Our relatively short irradiation near the parent ion frequency minimizes the size of the blind spot, but its almost complete absence is attributable to the collision delay following the SORI pulse. The blind spot is only produced during the 120 ms in which the SORI pulse is applied. Therefore, if a parent ion is acti- vated during the SORI pulse but does not dissociate

until the 1000 ms collision time, its product ions are not ejected. In other words, the use of a collision period fol- lowing SORI activation fills in the blind spot.

DISCUSSION

General structural implications

Non-linear pseudo-first-order behavior in gas-phase ion-molecule reactions may be caused by the presence of either multiple reactant ion structures or multiple product structures. In the current study, two types of ion-molecule reactions, deprotonation and H-D exchange, yielded kinetic data indicating that two ion populations are reacting at distinctly different rates. In addition, at least two different SORI CID spectra are seen. Deprotonation, H-D exchange and CID each gen- erate different product ions with their common feature being the reactant ions. Thus, it is reasonable to assume that the behavior observed here is the result of two or more different structures for the reactant ion, ubiquitin [M + 12H]12+.

Electrospray is a very soft ionization technique; therefore, these [M + 12H]12+ isomers do not result from rearrangements of atoms within the ubiquitin backbone. Instead, one possibility is that they are con- formational isomers with different three-dimensional structures. An alternative is that the two [M + 12H]"+ populations are protonated at different basic sites on ubiquitin. In addition, these types of isomers should not be viewed as mutually exclusive. The conformations of ubiquitin in solution may influ- ence the basic sites that are accessible to protonation during the ESI process. Conversely, the locations of the protonated sites may affect the intramolecular hydrogen-bonding in ubiquitin and thus impact its gas- phase three-dimensional structure.

Exposure to acid and organic solvents breaks the hydrogen-bonding linkages in ubiquitin leading to denatured ions that have unfolded relative to the native solution-phase three-dimensional Both acid and methanol were added to the ubiquitin solutions employed in this study. In addition, the [M + nH]"+ distribution of n = 7-13 that we observei6 is consistent with the distribution found for predomi- nantly denatured ubiquitin ion^.^*^* Therefore, a large percentage of the [M + 12H]12+studied here may have unfolded relative to their native aqueous-phase confor- mation.

For proteins and peptides, hydrogen-deuterium exchange reactions are generally used as probes of con- formational isomers. Labile hydrogens (i.e. those attached to oxygen and nitrogen) exchange more rapidly if they are readily accessible to the deuterating reagent and more slowly if they are buried in a fold of the protein.'1*32-34 The bimodal distribution observed for H-D exchange of [M + 12H]I2+ supports the present of conformation isomers. A rapid splitting of

252 C. J. CASSADY AND S. R. CARR

ions into a bimodal distribution during H-D exchange experiments is not common, but has recently been observed by Zhang and Smith.35 They have interpreted this distribution to mean that a two step process is occurring with the first step being protein unfolding and the second step being H-D exchange; if the rate of exchange is much greater than the rate of unfolding, ions that unfold earlier would surge ahead in the exchange process relative to species that had not yet unfolded. Although we cannot completely rule out this mechanism for ubiquitin [M + 12H]”+, it is unlikely to be the major process occurring because one of the nodes of the distribution can be removed prior to H-D exchange by selective deprotonation of the fast-reacting [M + 12H]’2+with ammonia. This suggests that the two populations of [M + 12H]‘2+ were present before the exchange reaction began. (Also, we observe the two populations by the independent techniques of deproton- ation reactions and CID.) In summary, the H-D exchange data suggest the presence of conformational isomers of [M + 12H]12+ with these isomers being gen- erated during the ESI process.

The presence of structural isomers resulting from dif- ferences in the sites of protonation is also consistent with our data, as is discussed below for the CID results. In addition, protonation site isomers would explain the unique behavior of [M + 12H]”+ relative to its neigh- boring ubiquitin ions. Assuming that the protons are located at the 13 most basic sites of ubiquitin, [M + 13H]13+ is fully protonated and will not have isomers related to protonation site. If two of the basic sites have roughly equal probability of being unprotonated, some [M + 12H]”+ ions may lack a proton at the first site while the others lack a proton at the second site. The result is two populations of [M + 12H]’*+ that are differentiated by protonation site and, as discussed above, by conformation. For [M + llH]’”, both of the two less favorable proto- nation sites would lack a proton with the result being one major ion population.

More specific structural implications from the CID and H-D exchange data

Support for the presence of [M + 12H]12+ protonation site isomers is provided by the CID data. In considering the CID spectra for [M + 12H]”+ produced directly from ESI, several notable changes occur as the fast- reacting ions are selectively removed by deprotonation reactions. For example, the abundances of y!S+ and y:: are highest when both [M + 12H]I2+ isomers are present at short reaction times but decrease as the fast- reacting ions are depleted. In contrast, production of yz;, y;:’ and y:: increases as the reaction progresses, meaning that these fragments are formed in greater abundance from slow-reacting [M + 12€1]12+. As can be seen from the ubiquitin structure in Fig. 5, the yss fragment contains amino acid residues 19-76, which have ten of the preferred basic sites; yZ4 contains resi- dues 53-76 with five basic sites. In the CID spectrum of fully protonated [M + 13H]I3+, yss has a maximum charge of 10+ and yZ4 has a maximum charge of 5+.

MQIFVKTLTGKTITLEVE PSDTIENVKAKIQDKEGIPPD

Y58

QQRLIFAGKQLED , GRTLSDY NIQKESTLHLVLRLRGG

Y24

Figure 5. Sequence of ubiquitin showing the locations of ya4 and yse cleavages. The most basic residues are given in larger, bold type.

Hence the preferential formation of yk:+ and y:: for slow-reacting [M + 12H]12+ suggests that these ions are protonated at all ten basic sites (or in their general vicinities) of residues 19-76 and all five sites of residues 53-76, respectively. For fast-reacting [M + 12H]I2+, the more abundant y!: supports loss of a proton from a basic site between residues 19 and 76, while the more abundant y:: suggests that a proton is absent between residues 53 and 76.

The indication from the CID data is that the slow- reacting, and presumably more energetically stable, [M + 12H]’2f lack a proton at a basic site near the N-terminus. Of the three basic sites in this region, the lysines of residues 6 and 11 are intrinsically more basic [GB of lysine = 218.7 kcal mol-’ (Ref. 36)] than the methionine of residue 1 [GB of methionine = 213.0 kcal mol-’ (Ref.3611. A recent ab initio study has shown that the N-terminal amino group of gas-phase peptide ions as small as protonated triglycine undergo extensive intra- molecular hydrogen bonding to remote parts of the mol- e ~ u l e . ~ ~ Therefore, it is possible that the N-terminal amino group for slow-reacting [M + 12H]”+ does not have its own proton, but may be sharing a proton with another basic residue via hydrogen bonding. In addi- tion, the slower rate of initial H-D exchange for this ion population suggests that it is more folded than the remaining [M + 12H]12+ ions. That is, this isomer may have retained more of ubiquitin’s native solution-phase structure.

For the fast-reacting, less stable ubiquitin [M + 12H]”+, the more facile initial H-D exchanges suggest a less folded structure relative to the structure of the other n = 12 population. In addition, the CID data for this isomer is consistent with lack of a proton on a basic site near the C-terminus. This segment of ubiqui- tin (residues 53-76) includes one lysine, one histidine and two arginine residues. Since arginine is by far the most basic amino acid in the gas phase [GB of arginine = 235.8 kcal mol-l (Ref.38)], intrinsic basicity considerations alone suggest that both arginine residues are protonated. It is possible that Coulombic repulsion is limiting protonation at one of these arginine residues because they are in close proximity (sites 72 and 74); however, this same reasoning would apply to the lysine residues of sites 27 and 29 where the CID data do not support structural differences. Another possible expla- nation is this [M + 12H]”+ structure is not proto- nated at the histidine residue of position 68. The relative basicities of histidine and lysine in the gas-phase are unclear. A recent s t ~ d y ~ ~ , ~ ~ has found that histidine has a gas-phase basicity that is -6 kcal mol-I lower than

ISOMERIC STRUCTURES OF UBIQUITIN IONS PRODUCED BY ESI 253

that of lysine, while other gas-phase report^^^,^'-^' show histidine to be slightly more basic by 0-2 kcal mol-'. It is well known, however, that in solution the side-chain of histidine is much more acidic than that of I y ~ i n e . ~ ~ Electrospray ionization generates the [M + 12H] l Z + ions from solution; therefore, solution- phase basicity may be limiting the protonation of histi- dine relative to that of the more basic lysine and arginine residues.

Correlation of proposed gas-phase structures with solid- and solution-phase data

Solid- and solution-phase data are consistent with the two structures of gas-phase ubiquitin ions proposed here. X-ray crystallography indicates that ubiquitin has an extremely compact structure that includes two mixed /?-sheet strands that are parallel and buried by other portions of the m ~ l e c u l e . ~ ~ - ~ ~ These contain residues 1-7 and 64-72, which incorporate the two regions that our CID data suggest contain unprotonated basic sites for the [M + 12H]"+ isomers. In solution, H-D exchange reactions followed by nuclear magnetic reso- nance detection have also revealed a tightly folded structure for native ~ b i q u i t i n . ~ ~ For the N-terminal methionine, which is proposed as lacking a proton in our slow-reacting isomer, crystallographic studies show that this residue is buried by two adjacent strands of /?-sheet and that its sulfur atom participates in hydro- gen bonding with the lysine on residue 63.44-46 In addi- tion, solution-phase alkylation reactions have found that the terminal methionine residue is buried and inac- cessible to reaction in the native aqueous-phase confor- mation but becomes exposed and reactive when conformational changes are induced by adding alcohol to the so l~ t ion .~ '

The C-terminal region of ubiquitin, which our CID data suggest may lack a proton for the fast-reacting [M + 12H]12+ isomer, also has interesting character- istics. In the native form, residues 64-72 are buried and generally i n a c ~ e s s i b l e ~ ~ - ~ ~ , but conformational changes are believed to occur in residues 64-76 that allow ubiquitin to undergo specific receptor interactions in living organism^.^^.^* In other words, during biological activity an initial site of unfolding is at the region near the C-terminus and, because this region aids in shield- ing the N-terminal residues, this unfolding also provides greater accessibility to the N-terminus.

Summary of structural implications

The two major ubiquitin [M + 12H]I2' isomers that we observe may be unfolded to different extents; this, in turn, may be either the cause or the result of different basic sites having access to protons. Of ubiquitin's 13 basic sites, the N-terminal amino group and the lone histidine residue are the least intrinsically basic. There- fore, these are logical sites to lack a protons in the two major [M + 12H]12+ structures. Given the distance between these sites (residue 1 versus residue 68), it is conceivable that the [M + 12H]"+ produced are dis- similar enough to have distinct reactivities and disso- ciation. It should be noted, however, that these are only proposed structures and that there are currently no techniques known to assign conclusively structures to multiply charged protein ions. In addition, other less abundant isomers, which cannot be distinguished by CID and ion-molecule reactions, may also be present.

CONCLUSION

Evidence for at least two forms of ubiquitin [M + 12H]12+ ions has been provided by deprotona- tion reactions, H-D exchange reactions and CID. Inter- estingly, similar studies on [M + 11H]"+ and [M + 13H]13+ have yielded no evidence for multiple structures. These ubiquitin species are one of the first multiply charged ion systems to be subjected to exten- sive structural studies in a mass spectrometer. There is a high probability that future research will discover iso- meric forms and unique chemistries for other gas-phase protein ions. The combination of techniques used here should provide a powerful tool for exploring structural problems for proteins relating to sequence, conforma- tion and protonation site.

Acknowledgements

The authors are indebted to Bruker Instruments for the loan of the ESI source and to G. H. Kruppa, J. Wronka, C. C. Stacey, C. H. Watson and F. H. Laukien for assistance or advice in utilizing the source. Financial support by the National Institutes of Health (R15- GM47657-01A1) and the Ohio Board of Regents Academic Challenge Program is gratefully acknowledged.

REFERENCES

1 .

2.

3.

4.

5.

C. M. Whitehouse, R . N. Dreyer, M. Yamashita and J. B. Fenn, Anal. Chem. 57,675 (1985). J. B. Fenn, M. Mann, C. K. Meng and S. F. Wong, Mass Specirom. Rev. 9. 37 (1 990). R. D. Smith, J. A. Loo, C. G. Edmonds, C. J. Barinaga and H. R . Udseth,Ana/. Chem. 62, 882 (1990). K . L. Duffin, J. K. Welply, E. Huang and J. D. Henion, Anal. Chem. 64,1440 (1 992). K. J. Light-Wahl, D. L. Springer, B. E. Winger, C. G. Edmonds, D. G. Camp, B. D. Thrall and R . D. Smith, J. Am. Chem. SOC. 11 5,803 (1 992).

6. R. Colton, B. D. James, I. D. Potter and J. C. Traeger, Inorg. Chem. 32,2626 (1 993).

7. S. K. Chowdhury, V. Katta and B. T. Chait, J. Am. Chem. SOC. 112, 9012 (1990).

8. J. A. Loo, R . R . Ogorzalek Loo, H. R. Udseth, C. G. Edmonds and R. D. Smith, Rapid Commun. Mass Spectrom. 5, 101 (1991).

9. U. A. Mirza, S. L. Cohen and B. T. Chait, Anal. Chem. 65, 1 (1993).

10. J. C. Y. Le Blanc, D. Beuchemin, K. W. Siu, R . Guevremont and S. S. Berman, Org. Mass Spectrom. 26, 831 (1 991 ).

254 C. J. CASSADY AND S. R. CARR

11. D. Suckau, Y. Shi, S. C. Beu, M. W. Senko. J. P. Quinn. F. M. Wampler and F. W. McLafferty, Proc. Natl. Acad. Sci. USA 90. 790 (1993).

12. T. D. Wood, R . A. Chorush, F. M. Wampler 111, D. P. Little, P. B. O’Connor and F. W. McLafferty, Proc. Natl. Acad. Sci. USA 92,2451 (1 995).

13. T. R. Covey and 0. J. Douglas, J. Am. SOC. Mass Spectrom. 4, 616 (1993).

14. K. A. Cox, R. K. Julian. Jr, , R . G. Cooks and R . E. Kaiser, Jr, J. Am. SOC. Mass Spectrom. 5,127 (1 994).

15. R . Feng and Y. Konishi, in Proceedings of the 47st ASMS Conference on Mass Spectrometry and Allied Topics, San Francisco, C 4 June 3, 1993, pp. 1020a-1020b (1 993).

16. C. J. Cassady, J. Wronka, G. H. Kruppa and F. H. Laukien, Rapid Commun. Mass Spectrom. 8, 394 (1 994).

17. P. Kofel, M. Allemann. H. P. Kellethals and K. P. Wanczek, lnt. J. Mass Spectrom. Ion Processes 65, 97 (1 985).

18. M. B. Comisarow, V. Grassi and G. Parisod, Chem. Phys. Lett. 57,413 (1 978).

1 9. J. E. Bartmess, in StructurelReactivity and Thermochemistry of Ions, edited by P. Ausloos and S. G. Lias, p. 367.Reidel. Dor- drecht (1 987).

20. J. E. Bartmess and R. M. Georgiadis, Vacuum 333, 149 ( 1 983).

21. J. W. Gauthier, T. R. Trautman and D. B. Jacobson, Anal. Chim. Acta 246, 21 1 (1 991).

22. M. Rechsteiner (Ed.), Ubiquitin. Plenum Press, New York (1 988).

23. J. A. Loo, C. G. Edmonds, H. R. Udseth and R . D. Smith, Anal. Chem. 62,693 (1 990).

24. A compilation of gas-phase basicities is given in S. G. Lias, J. F. Liebman and R . D. Levin, J. Phys. Chem. Ref. Data 13, 695 (1984); for consistency, all GBs discussed in the text are ref- erenced to this compilation.

25. T. Su and M. T. Bowers, lnt. J. Mass Spectrom. /on Phys. 12,

26. J. A. Loo, C. G. Edmonds and R. D. Smith,Anal. Chem. 65,

27. R. Bakhtiar, C. M. Holznagel and D. B. Jacobson,

28. M. W. Senko, J. P. Speir and F. W. McLafferty, Anal. Chem.

347 (1 973).

425 (1993).

Organometallics 12, 880 (1 993).

66,2801 (1 994).

29. S. A. Hofstadler, J. H. Wahl, R. Bakhtiar, G. A. Anderson, J. E. Bruce and R. D. Smith, J. Am. SOC. Mass Spectrom. 6, 894 (1 994).

30. V. Katta and B. T. Chait, Rapid Common. Mass Spectrom. 5, 214 (1991).

31. K. D. Wilkinson and A. N. Mayer, Arch. Biochem. Biophys. 250,390 (1 986).

32. D. L. Smith, 2. Zhang and Y. Liu, Pure Appl. Chem. 66, 89 (1 994).

33. V. Katta and 8. T. Chait,J. Am. Chem. SOC. 115, 6317 (1993). 34. R . J. Anderegg, D. S. Wagner, C. L. Stevenson and R. T. Bor-

chardt, J. Am. SOC. Mass Spectrom. 5,425 (1994). 35. 2. Zhang and D. L. Smith, presented at the 43rd ASMS Con-

ference on Mass Spectrometry and Allied Topics, Atlanta, GA, May 26,1995.

36. G. S. Gorman, J. P. Speir, C. A. Turner and 1. J. Amster, J. Am. Chem. SOC. 114,3986 (1 992).

37. K. Zhang, C. J. Cassady and A. Chung-Phillips, J. Am. Chem. SOC. 116,11512 (1994).

38. Z. Wu and C. Fenselau, Rapid Commun. Mass Spectrom. 6, 403 (1 992).

39. A. A. Bliznyuk, H. F. Schaefer and I. J. Amster, J. Am. Chem. SOC. 11 5,5149 (1 993).

40. Z. Wu and C. Fenselau, Rapid Common. Mass Spectrom. 8, 777 (1994).

41. G. B0jesen.J. Am. Chem. SOC. 109, 5557 (1 987). 42. K. Isa, T. Omote and M. Amaya, Org. Mass Spectrom. 25, 620

(1 990). 43. R . M. C. Dawson, D. C. Elliot, W. H. Elliot and K. M . Jones,

Data for Biochemical Research. Oxford University Press, Oxford (1 969).

44. S. Vijay-Kumar. C. E. Bugg, K. D. Wilkinson, R . D. Vierstra, P. M. Hatfield and W. J. Cook, J. Biol. Chem. 262, 6396 (1 987).

45. S. Vijay-Kumar, C. E. Bugg, K. D. Wilkinson and W. J . Cook, Proc. Natl. Acad. Sci. USA 82, 3582 (1 985).

46. S. Vijar-Kumar. C. E. Bugg and W. J. Cook, J. Mol. Biol. 194, 531 (1987).

47. P. D. Cary. D. S. King, C. Crane-Robinson, E. M. Bradbury, A. Rabbani, A. H. Goodwin, and E. W. Johns, Eur. J. Biochem. 11 2, 577 (1 980).

48. T. St. John, W. M. Gallatin, M. Siegelman, H. T. Smith, V. A. Fried and I. L. Weissman, Science 231, 845 (1 986).